Blockholder Trading, Market Efficiency, and Managerial Myopia

Transcription

1 THE JOURNAL OF FINANCE VOL. LXIV, NO. 6 DECEMBER 2009 Blockholder Trading, Market Efficiency, and Managerial Myopia ALEX EDMANS ABSTRACT This paper analyzes how blockholders can exert governance even if they cannot intervene in a firm s operations. Blockholders have strong incentives to monitor the firm s fundamental value because they can sell their stakes upon negative information. By trading on private information (following the Wall Street Rule ), they cause prices to reflect fundamental value rather than current earnings. This in turn encourages managers to invest for long-run growth rather than short-term profits. Contrary to the view that the U.S. s liquid markets and transient shareholders exacerbate myopia, I show that they can encourage investment by impounding its effects into prices. The nature of competition has changed, placing a premium on investment in increasingly complex and intangible forms the kinds of investment most penalized by the U.S. [capital allocation] system. Porter (1992) THIS PAPER ANALYZES how outside blockholders can induce managers to undertake efficient real investment through their informed trading of the firm s shares. By gathering information about a firm s fundamental value and impounding it into prices, they encourage managers to undertake investment that increases long-run value even if it reduces interim profits. The model therefore addresses two broad issues. First, it introduces a potential solution to managerial myopia. Second, it demonstrates that shareholders can add significant value even if they cannot intervene in a firm s operations. This may explain the prevalence in the Alex Edmans is at the Wharton School, University of Pennsylvania. I am deeply grateful to my advisors, Stewart Myers, Dirk Jenter, Gustavo Manso, and Xavier Gabaix, for their support and guidance. I am indebted to an anonymous referee for numerous comments that significantly improved the paper, as well as an associate editor and the editor, Cam Harvey. This paper has also benefited from input by Prasun Agarwal; Franklin Allen; Jack Bao; Indraneel Chakraborty; Florian Ederer; Isil Erel; Chris Evans; Carola Frydman; Itay Goldstein; Li He; Cliff Holderness; Adam Kolasinski; Pavitra Kumar; Robert Marquez; Anna Obizhaeva; Weiyang Qiu; Robert Ready; Michael Roberts; Steve Ross; Alan Schwartz; Jeremy Stein; Luke Taylor; Masako Ueda; Jialan Wang; Paul Yan; and seminar participants at the 2008 Penn/NYU Conference on Law and Finance, the 2007 WFA meetings, the 2006 CEPR European Summer Symposium in Financial Markets, the 2006 FMA doctoral tutorial, Berkeley, Boston College, Dartmouth, Duke, Maryland, MIT, Northwestern, Notre Dame, NYU, UBC, UCLA, UNC, Wharton, and Yale. I acknowledge excellent research assistance from Qi Liu, and travel grants from the NYSE and the MIT Graduate Student Council. This paper was previously circulated as Blockholders, Market Efficiency, and Managerial Myopia. 2481

2 2482 The Journal of Finance R United States of small transient blockholders, who typically lack control rights and instead follow the Wall Street Rule of voting with their feet selling their stock if dissatisfied. Many academics and practitioners believe that myopia is a first-order problem faced by the modern firm. In the last century, firms were predominantly capital-intensive, but nowadays competitive success increasingly depends on intangible assets such as human capital and R&D capabilities (Zingales (2000)). Building such competencies requires significant and sustained investment. Indeed, Thurow (1993) argues that investment is an issue of national importance that will critically determine the U.S. s success in global competition. However, managers may fail to invest if they are concerned with the firm s short-term stock price. Because the benefits of intangible investment are only visible in the long run, the immediate effect of such investment is to depress earnings. Indeed, Graham, Harvey, and Rajgopal s (2005) survey finds that 78% of executives would sacrifice long-term value to meet earnings targets. Moreover, the recent rise in equity-based compensation 1 (Murphy (2003)) and the sensitivity of CEO turnover to the stock price (Kaplan and Minton (2006)) has likely increased managers myopic tendencies. While previous papers have focused on various causes of myopia, this paper analyzes a solution: blockholders. A blockholder has strong incentives to gather costly information about the firm s fundamental value, that is, to learn whether weak earnings result from low firm quality or desirable long-term investment. 2 If the cause is low quality, the blockholder profits by selling her stake, depressing the stock price. If the cause is desirable investment, she does not sell, which attenuates the stock price decline caused by weak earnings. In both cases, the blockholder causes stock prices to reflect fundamental value rather than shortterm earnings. This increased market efficiency improves real efficiency: The manager is willing to undertake investments that boost fundamental value even if they depress short-term earnings. The Wall Street adage that the market sells first and asks questions later does not apply to blockholders: Owing to their sizable holdings, they have the incentive to ask questions first and not automatically sell upon losses. A noted real-life example is Warren Buffett, who typically acquires blocks in companies with significant growth opportunities. His investment shields the firm from stock market concerns, helping it to focus on long-term value. In sum, while the blockholder s monitoring and trading is motivated by her private desire to earn profits at the expense of liquidity investors, these actions have real social benefits by inducing efficient investment. But why does such monitoring and trading have to be undertaken by blockholders? Many empirical studies use block size as a proxy for investor sophistication, 1 Equity compensation would not induce myopia if it had very long vesting periods. However, vesting periods are often short in practice (Kole (1997), Johnson, Ryan, and Tian (2009)), perhaps because very long vesting periods would subject the manager to high risk. 2 Investment can encompass any action that enhances firm value, but worsens outsiders perceptions in the short run. Low investment can therefore represent accounting manipulation to improve outsiders short-term perceptions. Blockholders can deter such manipulation as they can see through the numbers and will sell if high earnings are not backed up by strong fundamentals.

3 Blockholder Trading, Market Efficiency, and Managerial Myopia 2483 assuming that large investors have greater incentives to become informed (see Rubin (2007) and Boehmer and Kelley (2009) for recent examples). Although this concept appears intuitive, it is not delivered in standard informed trading models because an investor s ability to trade on information is independent of her stake. The model introduces a short-sales constraint to provide a theoretical framework underpinning the assumed positive link between block size and information. If short sales are prohibited (or sufficiently costly), the only way to profit from bad news is to sell an existing position. The larger the initial holding, the more the blockholder can sell upon negative information, and so the greater the incentives to collect information in the first place. The above result requires the ability to solve for the blockholder s optimal trading volume upon information, as this determines how much information she will acquire. The Kyle (1985) model requires both normal liquidity trader demand and normal firm value to derive informed trades in closed form. Most corporate finance applications of the Kyle model use binary firm value and thus cannot solve for trading volumes endogenously; they therefore must restrict them to exogenous amounts that are then necessarily independent of initial holdings. This paper makes the methodological contribution of introducing exponential liquidity trader demand into a Kyle model to allow informed trades to be derived in closed form when firm value is discrete. It may be of use to future corporate finance applications of the Kyle model. The role of blockholders in this paper differs from prior models, where they add value through direct intervention, such as forcing a restructuring or vetoing a pet project. 3 This paper shows that blockholders can improve firm value even if they are unable to intervene (engage in voice ) and can only trade (engage in exit ). This departure from the literature is empirically motivated. Blockholders in the U.S. rarely intervene because they are typically small 4 and face significant legal and institutional barriers. 5 Existing models centered on control have difficulty in explaining the role that such non-controlling blockholders 3 Examples include Shleifer and Vishny (1986), Burkart, Gromb, and Panunzi (1997), Pagano and Röell (1998), Kahn and Winton (1998), Bolton and von Thadden (1998), Maug (1998, 2002), Aghion, Bolton, and Tirole (2004), Faure-Grimaud and Gromb (2004), and Brav and Mathews (2008). 4 When blockholders are defined as 5% shareholders, Holderness (2009) finds that 96% of U.S. firms contain a blockholder. However, when the minimum ownership is defined as 20%, La Porta, Lopez-de-Silanes, and Shleifer (1999) document that 20% (10%) of large (medium) U.S. firms contain a blockholder. They estimate that a 20% stake gives effective control if the shareholder is an insider; the threshold is likely to be higher for outside shareholders. Hence, blockholders are prevalent in the United States, but tend to lack control rights. Holderness finds that concentrated blockholders with board seats tend to be families. This paper focuses on financial blockholders who typically hold smaller stakes and have less frequent board representation. 5 See Becht et al. (2009) and Black (1990) for details of such barriers. Armour et al. (2007) document that U.S. shareholders seldom engage in litigation, and rarely succeed if they do; the same is true for the proxy fight mechanism. As Lowenstein (1988 p. 91) writes: [Institutional investors] implicitly praise or criticize management, by buying or selling, but seldom get involved more directly, even to the extent of a phone call. There is almost no dissent from the Wall Street Rule.

4 2484 The Journal of Finance R play in corporate governance, and thus justifying why they are so prevalent. 6 Moreover, the lack of large blockholders and consequent rarity of intervention may suggest that U.S. firms are poorly governed and that policy action is desirable. This paper offers an alternative perspective blockholders can still exert governance even if they lack control rights. In addition to evidence on the barriers to intervention, there is also research on the real-life importance of governance through trading. The survey evidence of McCahery, Sautner, and Starks (2008) finds that institutions use exit more frequently than any other governance mechanism, and Parrino, Sias, and Starks (2003) document direct evidence of this channel. The analysis of different governance mechanisms (exit vs. voice) leads to different results regarding the optimal block size for firm value and the effect of liquidity on governance. In a number of theories of voice, a larger block is always desirable as it increases monitoring and intervention incentives. Here, block size has a nonmonotonic effect on firm value. Trading profits depend not on block size per se, but on the amount sold upon bad news. If the block becomes too large, market liquidity declines and the blockholder cannot sell her entire stake. Because her potential trading profits are lower, she acquires less information, so prices are less efficient. This finite optimum is consistent with the paucity of large blockholders in the United States (La Porta, Lopezde-Silanes, and Shleifer (1999)). Large stakes may not only be unnecessary for a blockholder to exert governance, but also undesirable. The second difference from prior research concerns the desirability of liquidity. Motivated by several intervention theories, Bhide (1993) argues that because blockholders add value through voice, and voice and exit are mutually exclusive, liquidity is harmful as it allows a shareholder to leave rather than intervene. In this paper, the blockholder adds value through loyalty to a fundamentally sound firm that suffers weak short-term earnings. Loyalty and exit are similarly mutually exclusive and so it may seem that liquidity is again undesirable because it allows shareholders to sell easily upon weak earnings, causing managers to focus excessively on earnings. Indeed, Porter (1992) and Thurow (1993) argue that the U.S. s liquid markets deter long-run investment, potentially endangering the U.S. s future international competitiveness. They advocate policies to reduce liquidity and thus create unconditionally long-run shareholders who never sell. However, this paper shows that the mutual exclusivity of loyalty and exit paradoxically leads to complementarities between them. If a blockholder has retained her stake despite low earnings, this is a particularly positive indicator of fundamental value if she could easily have sold instead. In short, the power of loyalty relies on the threat of exit. The result that blockholders promote long-run behavior does not stem from simply assuming that blockholders are unconditionally long-run investors who never sell. Indeed, an investor s loyalty upon bad news is uninformative if market illiquidity prevented her from 6 Similarly, intervention models would imply little role for holders of nonvoting shares, even if they have large stakes. This paper shows that such investors can improve firm value by their trading behavior.

5 Blockholder Trading, Market Efficiency, and Managerial Myopia 2485 exiting anyway. Instead, conditional loyalty and the threat of short-term selling can, surprisingly, promote long-term investment. Far from exacerbating myopia, the liquidity of the U.S. capital allocation system may be a strength. This implication may explain why the above fears for the U.S. s global competitiveness have not been borne out. Consistent with the model, Fang, Noe and Tice (2009) document a causal relationship between liquidity and firm performance, which arises because liquidity leads to greater price efficiency. This result builds on Maug (1998), who overturned earlier papers arguing that liquidity is undesirable in an intervention setting by pointing out that they assume an exogenous block size. Maug shows that when block size is endogenous, liquidity is beneficial because it encourages a larger block to form in the first place. In the present paper, liquidity has a second benefit it leads to increased liquid trading by these blocks, once formed. While such trading is harmful in an intervention framework (and so the benefits of liquidity stem entirely from their effect on initial block size), in an exit model it is the very mechanism through which the blockholder adds value. Hence, increasing liquidity from low levels is beneficial even if block size is exogenous and does not rise in response to greater liquidity. 7 Here, liquidity is desirable not only because it leads to larger blocks (Maug s result), but also because it leads to more liquid trading by these blocks. This conclusion is potentially important because disclosure requirements or regulation may restrict block size from rising in response to liquidity (see, for example, Roe (1994)) and prevent the first benefit from being obtained. Indeed, in the United States, ownership is fragmented despite high liquidity, suggesting that these forces may be important. Even if this is the case, liquidity can be valuable. While the paper s main result is that blockholders can encourage investment, the corollary is that a key cost of the U.S. s dispersed ownership is myopia. This leads to an additional policy implication. Previous papers argue that the main problem with atomistic shareholders is that they lack the control rights to intervene and thus allow the manager to shirk (e.g., Roe (1994)). In this case, potential solutions to dispersed ownership are equity compensation and regulations against takeover defenses. However, if the main cost is that dispersed shareholders lack information on fundamental value, focus on current earnings, and thus induce myopia, these policies exacerbate the issue. The problem with small shareholders may not be so much the separation of ownership from control (Berle and Means (1932)) as the separation of ownership from information. The paper closes with empirical implications. One set concerns stock price effects, and is unique to a model in which blockholders trade rather than intervene. While block size does not matter in standard microstructure theories, here it is positively correlated with an investor s private information, trading profits, and price efficiency. Existing empirical studies typically use institutional ownership as a measure of investor informativeness, but the model suggests 7 When block size is exogenous, the optimal liquidity is finite because too much liquidity camouflages the blockholder s trades and reduces price informativeness. With endogenous stakes, increasing liquidity is desirable even at high levels because block size increases with liquidity to prevent such camouflage.

6 2486 The Journal of Finance R that block size may be more relevant, because concentrated stakes are necessary to incentivize investors to gather information. Bushee and Goodman (2007) and Parrino, Sias, and Starks (2003) indeed find that larger shareholders are more informed. More generally, the model suggests a shift in focus in the way we think about blockholders that can give rise to new directions for empirical research. Previous studies have been primarily motivated by perceptions of blockholders as controlling entities and thus analyze private benefits (e.g., Barclay and Holderness (1989, 1991)), but new research questions may be motivated by conceptualizing them as informed traders and thus analyzing their effect on financial markets. (See Gallagher, Gardner, and Swan (2008) and Brockman and Yan (2009) for two recent such papers.) A second set relates to real effects: Blockholders should increase firm investment (Cronqvist and Fahlenbrach (2009)), and deter earnings manipulation (Dechow, Sloan, and Sweeney (1996), Farber (2005), Burns, Kedia, and Lipson (2008)). These predictions distinguish the model from theories where the blockholder solves managerial shirking or the pursuit of pet projects, rather than myopia. Admati and Pfleiderer (2009) also analyze a blockholder who can only engage in exit rather than voice. Our papers focus on fundamentally different agency problems: While Admati and Pfleiderer analyze effort (broadly defined to incorporate both shirking and free cash flow problems), I study investment, in particular, whether it is deterred by liquid trading as commonly believed. While the effort conflict may have been attenuated by recent increases in the manager s sensitivity to the stock price, such changes further exacerbate myopia. A second difference is that Admati and Pfleiderer assume that the blockholder is exogenously informed, and so the level of monitoring is fixed. This paper endogenizes costly information gathering and generates testable predictions regarding the effect of block size on monitoring and trading, and in turn market efficiency, real investment, and firm value. By contrast, Admati and Pfleiderer focus on the nature of the agency problem and derive the interesting result that while the blockholder always attenuates free cash flow problems, she sometimes exacerbates shirking. The beneficial effect of ex post monitoring on ex ante investment is shared by Edmans (2009a). Debt concentrates equityholders stakes, incentivizing them to discover the cause of interim losses. Thus, debt can allow liquidation of an incompetent manager who suffers short-term losses, without simultaneously deterring skilled managers from long-term projects that risk such losses. While Edmans (2009a) is a theory of capital structure that assumes intervention, this paper is a theory of ownership structure in which the blockholder can only trade. In Holmstrom and Tirole (1993), increased market efficiency allows prices to more accurately reflect managerial effort, increasing the optimal sensitivity of pay to the stock price. In their model, monitoring is performed by atomistic shareholders and they do not consider blockholders as potential monitors hence, blocks reduce price efficiency via their negative effect on liquidity. I show that blockholders may be particularly important monitors as they have the strongest incentives to gather the intangible information that is especially relevant for long-term investment. Without blockholders, the stock price primarily

7 Blockholder Trading, Market Efficiency, and Managerial Myopia 2487 t=1 Public signal s {s g, s b } released t=2 Blockholder exerts monitoring effort µ, receives signal i {i g, i b }, and sells b [0, ] Liquidity traders demand u ~ exp(λ) Market maker observes d = b + u and sets P = E[V d, s] t=3 V is publicly revealed Figure 1. Timeline of the model. reflects publicly available current earnings. Thus, tying the manager s pay to the firm s stock price can induce myopia. A final strand of related literature concerns insider trading by management, which can also increase financial and real efficiency (e.g., Manne (1966)). The blockholder is likely to be significantly more effective than the manager at impounding information into prices for several reasons. First, managers are constrained by insider trading laws, personal wealth (limiting purchases), or lock-ups of stock as part of incentive packages (limiting sales). Second, the manager may be conflicted because the stock price is used to evaluate him, and so may choose not to reveal negative private information by selling shares. Third, conflicts may also arise because the manager has control over the information flow and investment decisions (Bernhardt, Hollifield, and Hughson (1995)). He may release false negative (positive) information and subsequently buy (sell) shares, or sell his shares and take the incorrect investment decision. One paper that does analyze insider trading by the blockholder is Maug (2002), who shows that legalizing such actions can induce her to exit rather than engage in value-enhancing intervention. This paper is organized as follows. Section I introduces the basic model, which links block size to financial efficiency. Section II presents the core result of the paper by introducing managerial decisions and illustrating the impact on real efficiency. Section III discusses empirical predictions and Section IV concludes. The Appendix contains all proofs not in the main paper. I. Blockholders and Market Efficiency This section analyzes the effect of block size on monitoring and stock prices. The real consequences are studied in Section II, where managerial decisions are introduced. I consider a firm with one share outstanding. A blockholder (B) owns units and atomistic shareholders collectively own the remaining 1 units. All agents are risk-neutral and the risk-free rate is normalized to zero. There are three periods, summarized in Figure 1. At t = 1, a public signal s {s g, s b }, such as an earnings announcement, is released. The signal is imperfectly informative about the firm s fundamental value V, which is revealed at t = 3. If

8 2488 The Journal of Finance R s = s g, V = X > 0 with certainty; if s = s b, V = 0orX with equal probability. I refer to a firm with V = X(0) as a high (low)-quality firm; s = s g is a good signal and s = s b is a bad signal (also referred to in the text as losses or low earnings ). At t = 2, B exerts monitoring effort µ [0, 1] at cost 1 2 cµ2. Monitoring gives B a private signal i {i g, i b } of V, the precision of which rises with µ as follows: Pr(i g X ) = Pr(i b 0) = µ Pr(i g 0) = Pr(i b X ) = µ. The posterior probabilities that the firm is of high quality are thus given by Pr(X i g ) = 1 + µ 2 Pr(X i b ) = 1 µ (1) = π b. 2 If µ = 0, private information is completely uninformative and the posterior equals the prior 1 ;ifµ = 1, B knows V with certainty. There is then a round 2 of trading. The blockholder either demands nothing (b = 0) or sells β units (b = β); she sells if she receives signal i b and holds otherwise. 8 I assume β owing to short-sales constraints because this paper s focus is noninterventionist financial blockholders such as mutual funds, pension funds, and insurance companies, the vast majority of which are unable to sell short. The model s results continue to hold under nontrivial short-sales costs, as discussed in Section I.A. Also at t = 2, liquidity traders demand u, where u is exponentially distributed, that is, { 0ifu 0 f (u) = λe λu if u > 0, where λ = 1 and ν 1 is a liquidity parameter. The competitive market ν(1 ) maker sees total demand d = b + u and sets a price P equal to the conditional expectation of V given d and s, similar to Kyle (1985). The parameter ν captures factors other than free float (1 ) that affect liquidity, such as transaction costs, taxes, disclosure requirements, and other regulations. Since the mean liquidity trade is E(u) = 1, we have the standard λ 8 The core analysis involves the blockholder selling or holding, because this paper s focus is the Wall Street Rule: the shareholder exit that is widely believed to exacerbate myopia. The results are unchanged by allowing the blockholder to buy a fixed amount regardless of her initial stake. While the incentives to buy are unaffected by, the ability to sell remains (nonmonotonically) increasing in. Hence, overall profits from information, and thus monitoring incentives, remain nonmonotonically increasing in. The results are in the Internet Appendix, available at

9 Blockholder Trading, Market Efficiency, and Managerial Myopia 2489 feature that the volume of liquidity trades is increasing in the amount held by small shareholders (1 ), because liquidity trades often emanate from current investors. While liquidity trades are literally modeled as purchases, all of the model s results continue to hold if the distribution of liquidity trades is transposed downwards, so that the bulk of such trades are sales. Because only current shareholders can sell the stock, there is a clear connection between free float and liquidity. This linkage remains under the literal interpretation of liquidity trades as purchases. Current shareholders are likely to be more informed about a particular stock than nonshareholders, and thus more likely to be purchasers if there is ambiguity aversion, or if nonshareholders do not know about the stock s existence see Merton (1987) for a model in which investors are restricted to buy stocks that they know about. In Bolton and von Thadden (1998), Kahn and Winton (1998), and Holmstrom and Tirole (1993), liquidity purchases also stem from existing owners. The exponential distribution of u is a methodological step that allows B s sale volume to be derived in closed form. 9 (The key idea that block size affects the ability to sell on negative information, and thus monitoring incentives, does not depend on the functional form for u). Kyle (1985) achieves tractability with normal liquidity trader demand as firm value is also normal. In most corporate finance models featuring the Kyle model, firm value is binary and so the informed trader s order cannot be solved for; such papers therefore typically restrict her trade to exogenous amounts. The exponential distribution in this paper allows her trade to be endogenously derived as a function of block size. A. Market Equilibrium If signal s g is emitted, the market maker knows that the firm is of high quality, and so sets P = X. Because the signal is fully revealing, B has no incentives to monitor or trade. The remainder of this section focuses on the interesting case of s = s b, and so s b notation is omitted for brevity. Because the signal is not fully revealing, B does monitor and trade, and the market maker tries to infer B s information from total order flow d. Lemma 1 below presents the Nash equilibrium, where B s trading and monitoring decisions are optimal given the market maker s pricing function, and the market maker s pricing function earns him zero profit given B s decisions. I assume X 8c to ensure that effort does not exceed the maximum of one. LEMMA 1: Upon observing s b and total demand d, the market maker sets the following prices: { P = πb X if d 0 P = π m X if d > 0, (2) 9 Barlevy and Veronesi (2000) also use an exponential distribution, but not to solve for trading volumes since informed traders always trade the maximum amount in their model.

10 2490 The Journal of Finance R where π m = Pr(X d > 0) = 1 + e λβ + µ(1 e λβ ). (3) 2(1 + e λβ ) The blockholder exerts monitoring effort µ = β X 4c. (4) If and only if she observes signal i b, B sells ( ) 1 β = min λ,. (5) A full proof is in the Appendix; here I summarize the key intuition. If d 0, the market maker knows that B has sold and thus has received i b. He therefore sets prices according to the posterior π b = Pr(X i b ) in equation (1). On the other hand, d > 0 is consistent with both selling and not selling. In this case the market maker sets prices according to the posterior π m = Pr(X d > 0) in equation (3). This gives rise to equation (2). If B receives signal i b, she wishes to sell. As in Kyle (1985), in the absence of short-sale constraints, her optimal trade is finite ( 1 ) as she is concerned with λ excessive price impact. With short-sale constraints, B is unable to sell more than, her initial holding. If 1, then β = : Liquidity is sufficiently high λ that B finds it optimal to sell her entire stake. LEMMA 2: The maximum sale volume β is given by β = = ν ν + 1. (6) Blockholder effort µ is also maximized when =. Both β and µ are increasing in if <, and decreasing in if >. For <, a larger initial stake raises the amount that B can sell upon negative information, and thus the incentives to become informed in the first place. Simply put, the benefits of information are higher as B can make greater use of it. Empirical studies frequently assume that incentives to monitor and trade are increasing in block size. 10 While intuitive, such a result is not delivered by standard models of informed trading with no constraints (e.g., Kyle (1985) and its variants). In these models, monitoring is independent of initial holdings: If an investor uncovers negative information about a stock she does not own, she can short sell. Similarly, in typical applications of the Kyle model to corporate finance, the informed party s trade is restricted to exogenous amounts and thus independent of her stake. These restrictions arise either from assumed trading or wealth constraints (e.g., in Boot and Thakor (1993) and Fulghieri and Lukin 10 For example, Boehmer and Kelley (2009 p. 3590) assume that a given institution will only acquire information on large ownership stakes. Rubin (2007 p. 224) posits that the probability that a particular institution will incur the costs to do so is higher if it enjoys a comparatively large ownership share.

11 Blockholder Trading, Market Efficiency, and Managerial Myopia 2491 (2001), the investor can only trade one unit), or because liquidity trades are discrete, forcing the informed party to match liquidity traders volumes to avoid being revealed (e.g. Maug (1998), Bolton and von Thadden (1998), Faure-Grimaud and Gromb (2004)). This paper generates a link between and µ via the combination of continuous liquidity trader demand, endogenous informed trader demand, and short-sales constraints. It thus provides a theoretical framework underpinning the above empirical assumption. 11 However, a second consequence of a higher stake is that it reduces liquidity. If >, liquidity is sufficiently low that B chooses to sell only 1 if she receives λ signal i b. Further increases in reduce liquidity and thus the optimal trading volume 1 ; because B expects to trade less on information, she has fewer incentives to gather information. The optimal block size to maximize information λ acquisition is therefore finite at. If B can short-sell at a cost (as is the case for hedge funds), it remains the case that increasing from zero augments β and µ, as long as the cost is sufficiently nontrivial that the reduction in short-sale costs that results from raising outweighs the negative effect on liquidity; the results are in the Internet Appendix. A higher initial stake increases the profits from selling on private information, because B can costlessly unwind a long position rather than engage in costly short-sales. Hence, B has greater incentives to acquire private information. As in the core model, once is sufficiently large, further increases in lower β and µ because the negative effect on liquidity dominates, so the relationship is again concave. Equity analysts are also potential monitors and can move prices without trading. The activity of equity analysts (and other hedge funds) is captured in the parameter c. This is B s cost of acquiring private information not already in the market and is therefore inversely related to the firm s information asymmetry. If analyst and hedge fund activity is high, most value-relevant information is already in the market price and the cost of acquiring incremental information is large. Section II demonstrates that this reduces the blockholder s value added. It is straightforward to show that B does not sell (hold) upon receiving i g (i b ). Selling in the absence of negative private information would drive the price down and reduce her portfolio value at t = 2 as well as t = 3. Hence, even a blockholder concerned with interim performance (e.g., a fund manager evaluated by investors) will not sell purely on public information. B. Market Efficiency This section analyzes the relationship between block size and price efficiency. A high-quality firm has a ( µ) chance of emitting signal i g, in which case B does not sell and the price is π m X. It has a ( µ) chance of emitting signal 2 i b, in which case B sells. If u β (which occurs with probability 1 e λβ ), then 11 In Van Nieuwerburgh and Veldkamp (2009), monitoring does increase in the investor s holding, but because she is risk-averse and wishes to reduce uncertainty, rather than because a larger block expands the set of feasible trading strategies.

12 2492 The Journal of Finance R d 0 and the price is π b X. Otherwise the price is π m X. Hence, the expected price of a high-quality firm is E[P X ] = X π X, where the expectation is taken over the possible realizations of i and u, and [ 1 π X = ( 1 2 µ ) ] ( 1 2 µ e λβ π m ) 2 µ (1 e λβ )π b = 1 (µ 2 1 ) e λβ e + 1. (7) λβ I use π X as a measure of market efficiency as it captures the closeness of expected prices to fundamental value. 12 If π X = 1, price equals fundamental value and the market is fully efficient. As π X declines, the expected price of the highquality firm falls from its fundamental value of X. PROPOSITION 1 (Market Efficiency): Market efficiency π X is maximized at =. It is increasing in for <, and decreasing in for >. Proof: If <, then β =. Differentiating equation (7) with respect to gives π X = µ2 λe λ (1 + e λ ) } {{ 2 } trading effect + µ 2 λ 1 e λ + µ 1 e λ µ. (8) (1 + e λ ) } {{ e } } {{ λ } camouflage effect effort effect The trading effect is the direct impact of. It is positive if and only if < because an increase in raises the amount sold by B upon negative information. Simply put, if B trades more, her trading (or nontrading) impounds more information into prices. The camouflage effect operates indirectly through decreasing liquidity. Because liquidity camouflages B s trades, this effect is positive for all levels of, as a fall in liquidity increases her effect on prices. The effort effect operates indirectly through affecting µ. This effect is positive if and only if µ > 0, that is, <. Increased effort leads to B receiving a more informative signal. Her trades thus convey greater information about V. Overall, if <, all three effects are positive, and so an increase in raises market efficiency. If >, then β = 1. Differentiating with respect to gives λ π X e 1 µ = µ1. (9) } 1 + {{ e 1 } effort effect 12 Note that the price is always efficient in the sense of equaling fundamental value conditional upon an information set. However, when rises, this information set is richer and so prices are closer to (unconditional) fundamental value.

13 Blockholder Trading, Market Efficiency, and Managerial Myopia 2493 From Lemma 2, the trading effect is negative, as liquidity is sufficiently low that B only sells 1, which is decreasing in. The negative trading effect exactly λ cancels out the positive camouflage effect. This leaves the effort effect, which is negative from Lemma 2. Q.E.D. Even considering only the benefits of blockholders and ignoring their costs, the optimal block size for market efficiency is a finite level,. (Section II shows that the that maximizes market efficiency also optimizes firm value.) This result contrasts with some intervention models such as Shleifer and Vishny (1986), Maug (1998), and Kahn and Winton (1998), where firm value is monotonically increasing in block size. 13 In this model, it is not block size per se that matters, but the associated optimal trading volume: Prices are a function not of, but of min( 1, ). A large block increases information revelation λ only to the extent that there is sufficient market liquidity to allow it to be sold entirely. Put differently, the fact that B has not exited is less of a positive boost to the stock price if exit was difficult in the first place. This finite optimum is consistent with the finding that, while blockholders are common in the United States (Holderness (2009)), substantial blockholders are rare (La Porta, Lopezde-Silanes, and Shleifer (1999)). In reality, other market participants may be able to observe blockholders sales with a lag by studying Section 13 filings. This would strengthen B s impact on market efficiency. Because sales are only observed with a lag, B s profits from informed selling are unchanged. However, her price impact is greater: After the filing is made, the price moves even closer to fundamental value because the market can now observe the trade directly. II. Blockholders and Long-Term Investment The previous section links blockholders to increased financial market efficiency. This section demonstrates that the latter can in turn augment real efficiency by addressing the potentially important myopia issue. I thus illustrate a social benefit for information gathering that is motivated purely by the private desire to profit from informed trading. The model is extended to allow for managerial decisions. The risk-neutral 14 manager (M) places weight ω on the t = 2 stock price and 1 ω on the t = 3 firm value, where 0 <ω<1. Because this paper focuses on the solution to myopia rather than its cause, the concern with the current stock price (ω >0) is taken as exogenous. This is a standard assumption in the literature and can 13 Holmstrom and Tirole (1993) and Bolton and von Thadden (1998) also derive a nonmonotonic effect of block size. In their models, market efficiency is maximized with a zero block. They derive finite optimal block sizes as they trade off market efficiency against, respectively, monitoring costs and intervention. In this paper, the optimal block size is finite even focusing on market efficiency alone. Burkart, Gromb, and Panunzi (1997) derive a finite optimal block size as too large a block can erode managerial initiative. In Pagano and Röell (1998), too large a block can lead to overmonitoring. 14 Introducing managerial risk aversion would strengthen the results because the blockholder reduces the variance in the price of a high-quality firm that emits s b, as well as increasing its mean.

14 2494 The Journal of Finance R be motivated by a number of underlying factors, such as takeover threat (Stein (1988)), concern for managerial reputation (Narayanan (1985), Scharfstein and Stein (1990)), or the manager expecting to sell his own shares at t = 2 (Stein (1989)). 15 At t = 0, the manager of a high-quality firm can invest in a long-term project that unambiguously increases fundamental value, but risks low interim earnings. The most natural example is intangible investment that is expensed and thus difficult to distinguish from losses made by a low-quality firm. Let θ [0, 1] denote the amount of investment. Investment of θ boosts the firm s t = 3 value to V = X + gθ, but risks emitting s b at t = 1 with probability θ 2 (otherwise, s g is emitted). The parameter g measures the productivity of the investment project. The choice of θ does not involve a personal utility cost to M: There is no standard effort conflict. The availability of this investment project is not known to the market maker nor to B, to emphasize the fact that B can induce M to exploit growth opportunities even if she is unaware of their existence (in contrast with intervention models). In reality, new investment opportunities that were previously unforeseen frequently become available to managers. The Internet Appendix shows that the results continue to hold if θ is anticipated. At t = 0, the manager of a high-quality firm chooses θ to maximize (1 ω)(x + gθ) + ωθ 2 π X X + ω(1 θ 2 )X. (10) The first term is firm value, multiplied by its weight in the objective function. The stock price is X if s g is emitted, which occurs with probability (1 θ 2 ), else π X X. This gives rise to the second and third terms. LEMMA 3: The manager chooses investment level θ given by ( ( ) ) 1 ω g θ = min 2ωX ( ),1. (11) 1 π X If θ<1, it is increasing in g and π X, and decreasing in X and ω. The amount of long-term investment is naturally increasing in its productivity g, and decreasing with the cost of emitting s b. The latter is positively related to the difference in value between high- and low-quality firms X, and M s concern for the current stock price ω. Note that myopia is rational: The stock price falls upon s b because it may have been emitted by a low-quality firm; given the risk of this decline, M optimally sets θ below its first-best level of one. Investment increases with π X because greater market efficiency means that prices more closely reflect fundamental value. Because π X in turn depends on, investment depends on block size. Indeed, taking first-order conditions of equation (11) with respect to and calculating cross-partials with respect to g, c, and ω leads to Proposition 2 below, the main result of the paper. 15 Even if the manager s sole objective is to maximize long-run shareholder value, he will care about the stock price as it affects the terms at which the firm can raise equity at t = 2 (Stein (1996)).

15 Blockholder Trading, Market Efficiency, and Managerial Myopia 2495 PROPOSITION 2 (Investment): Define X 1 = (1 ω)g and X ω 2 = (1 ω)g 1+e 1. For all ω 2e 1 X, investment θ is weakly increasing in π X. It is therefore maximized at =, weakly increasing in for <, and weakly decreasing in if >. If X X 2, these directional effects are strict and θ is uniquely maximized at =. If X X 1, M invests efficiently (θ = 1) regardless of π X and thus. The magnitude of the block-sensitivity of investment θ is weakly increasing in g and weakly decreasing in c and ω, that is, 2 θ g 0, 2 θ c 0, and 2 θ ω 0 for < and 2 θ g 0, 2 θ c 0, and 2 θ ω 0 for >. The central result of this paper is that blockholders can add value, even in the absence of an underlying effort conflict and the ability to intervene. By engaging in informed trading to maximize their own speculative profit, they can promote long-term investment. For <, increasing block size raises market efficiency π X (Proposition 1) and thus makes the price more closely reflect fundamental value. From equation (11), a higher π X in turn augments real efficiency: The manager is more willing to undertake positive-npv longterm investment projects that risk interim turbulence because the stock price fall upon short-term losses is attenuated. While Proposition 2 shows that sizable shareholders promote investment, the corollary is that a key cost of dispersed ownership is that it magnifies myopia. This cost contrasts with the shirking traditionally focused upon (e.g., Roe (1994)) and has different policy implications. If effort is the main problem, equity compensation and a more active takeover market are potential solutions. However, if myopia is the principal issue, such measures make it worse. The beneficial effect of a blockholder on investment, θ, is decreasing in c and thus increasing in information asymmetry. Where information asymmetry is high, there is more information for B to impound into prices so she has a greater incremental effect. The blockholder s impact is also increasing in the profitability of investment g up to a point ( 2 θ > 0): If the investment is unattractive, g it will be little exploited even if B makes prices relatively efficient. However, if g is sufficiently high that X X 1, θ = 1 and θ = 0: The investment opportunity is sufficiently attractive that M pursues it fully even in the absence of a blockholder. In a similar vein, the impact of higher block size is greatest for moderate levels of ω. IfM is greatly concerned with interim performance, he will still underinvest even in the presence of a blockholder ( 2 θ < 0). On the ω other hand, if the stock price is a minor concern (X X 1 ), M invests efficiently in the first place. A. Does Liquidity Deter Investment? The previous section studies the optimal for firm value, holding liquidity constant. This section now examines the effect of liquidity ν on investment. I first assume that is exogenous and show that increasing ν boosts investment at low levels, but reduces it at high levels. Next, I allow to be endogenously chosen by the blockholder in response to liquidity, in order to maximize her total payoff. In this case, increasing ν always boosts investment.

16 2496 The Journal of Finance R A.1. Exogenous Block Size PROPOSITION 3 (Liquidity, Exogenous Block Size): Holding constant, market efficiency and investment are maximized at ν =. They are increasing (decreasing) in ν for ν<(>) ν 1. From Proposition 2, investment is increasing in market efficiency. Market efficiency in turn depends on two factors: how much information B gathers, and the extent to which this information is incorporated into prices. While liquidity increases information gathering by augmenting trading profits, it also camouflages B s trades and reduces their price impact. For low (high) levels of efficiency, the first (second) effect dominates. If there is zero liquidity, B does not trade or monitor; if liquidity is infinite, she does not affect prices. The nonmonotonic effect of liquidity contrasts with previous papers such as Holmstrom and Tirole (1993) and Faure-Grimaud and Gromb (2004), where augmenting liquidity always increases stock price informativeness. There is no camouflage effect in those papers as the informed investor s trades are unbounded; here, B s maximum sale is capped at due to short-sales constraints. 16 As summarized by Bhide (1993), liquidity is undesirable in most previous papers, where block size is exogenous and the blockholder chooses between intervention and intentional exit. 17 In such papers, the blockholder adds value through voice; because voice and exit are mutually exclusive, liquidity hinders the former by facilitating the latter. Here, the blockholder adds value by retaining her stake through interim turbulence, increasing investment ex ante. Because loyalty and exit are similarly mutually exclusive, it might seem that liquidity is again undesirable as it encourages exit and thus deters loyalty. This is indeed the conventional wisdom: Liquidity allows shareholders to sell upon weak earnings, and thus makes managers even more concerned with earnings. A number of commentators (e.g., Porter (1992), Thurow (1993)) argue that the U.S. s liquid capital markets hinder long-term investment, and hence have called for policy intervention to reduce liquidity. This paper shows that, even holding exogenous, increasing liquidity from low levels can promote investment, and thus has very different policy implications. Although loyalty and exit are indeed mutually exclusive, this leads to complementarities between them. The power of loyalty relies on the threat of exit. By making exit more feasible, increased liquidity renders loyalty more meaningful. In this model, the blockholder does not promote investment simply by being a long-term investor who never sells; by contrast it is the possibility 16 Some previous blockholder models (where is not chosen by B) also conclude liquidity is not unambiguously desirable. In Kahn and Winton (1998), liquidity has no effect, rather than a nonmonotonic effect. Bolton and von Thadden (1998) do feature an optimal level of liquidity. This arises because greater liquidity means a lower stake: In their paper, liquidity is (1 ), so higher liquidity can only be achieved by a lower, which reduces intervention and thus firm value. In this paper, liquidity is ν(1 ), where ν captures factors that affect liquidity unrelated to free float. The model shows that there is an optimal ν, even if is constant. 17 Faure-Grimaud and Gromb (2004) demonstrate that liquidity encourages intervention as it allows the stock price to reflect these value gains and thus the blockholder to earn a return if she has to exit unexpectedly, due to a liquidity shock. In their model, exit is not intentional.

17 Blockholder Trading, Market Efficiency, and Managerial Myopia 2497 of selling in the short-run that encourages the manager to make long-term decisions. Indeed, if market illiquidity compelled the blockholder always to hold for the long run, she has no effect on stock prices and investment. The fact that she has not sold upon bad news is uninformative if she was unable to sell in the first place. This result marks an important distinction from intervention models. If the blockholder has no control rights, allowing her to sell in the short term is beneficial for firm value as it can promote investment. By contrast, if the blockholder is interventionist, the possibility of short-term selling may induce her to step in and force the manager to undertake myopic decisions. Therefore, not only is it unnecessary for blockholders to have control rights in order to exert governance, but it may also be undesirable: To the extent that blockholders have short-term considerations, they may add more value to the firm if they lack control rights. A short-term blockholder without control rights might try to induce myopia by threatening to punish the manager by selling her stake if earnings are low (because of investment). However, such a threat is not credible as it is dynamically inconsistent: Once the firm announces low earnings, they are immediately incorporated into the stock price, and so the blockholder cannot profit by selling. Opponents of liquidity argue that it allows shareholders to sell upon interim losses, but this view lacks a theoretical framework to explain why investors would sell upon weak earnings. In an efficient market, the stock price reacts immediately to public information such as low earnings, removing the incentive to exit. Since investors can only profit by trading on private information, trading is desirable as it impounds such information into the stock price particularly if the trader has a sizable stake and so is likely to have engaged in fundamental analysis. Indeed, Yan and Zhang (2009) show empirically that investors who trade frequently are better informed than those who rarely trade. Moreover, the stocks that they own do not exhibit long-run reversals, which is inconsistent with the view that they encourage short-termism. The frequent trading observed in the United States may thus be a positive sign, as it suggests that information is being impounded into prices. A.2. Endogenous Block Size Thus far, the analysis has focused on the normative issue of the block size that maximizes firm value,. I now turn to the positive question of which block size is most likely to be observed empirically. While the firm may be able to influence temporarily (e.g., by privately placing equity with a blockholder), B can trade away from this initial stake. The only initial block size that is robust to re-trade is the private optimum, P, that maximizes B s trading profits net of monitoring costs. This would also be the block size that B would choose if she bought shares at t = 0 and her purchase was unobserved. 18 (After the purchase, becomes publicly known through Section 13 filings.) 18 In Kyle and Vila (1991), the blockholder camouflages her purchase by trading with liquidity investors. In their model, this leads to an additional benefit of liquidity it facilitates initial block acquisition.

18 2498 The Journal of Finance R I now characterize this private optimum. If B owns, her expected gross trading profits are given by () = Pr (s b) 1 2 βe λβ X (π m π b). The first term is the probability that s b is emitted 19 and the second is the probability that B receives signal i b, in which case she sells β. With probability e λβ, u >β and so B s information is not revealed to the market; she earns X(π m π b ) per unit. Her expected monitoring costs are given by () = Pr (s b) β2 X 2 32c, and so her objective function is () () = Pr (s b) X 2 ( β 2 e λβ ). (12) 8c 1 + e 1 λβ 4 Proposition 4 below states that both P and θ are monotonically increasing in ν. PROPOSITION 4 (Liquidity, Endogenous Block Size): The privately optimal block size P is strictly less than the firm value optimum, and monotonically increasing in liquidity ν. Allowing for the endogeneity of block size, investment θ is monotonically increasing in ν. Two forces reduce P below. First, a larger block size augments monitoring costs, which the blockholder bears but are absent from the firm s objective function. Second, a larger block reduces liquidity and thus trading profits, which are also absent from the firm s objective function. By contrast, a decrease in liquidity has a direct positive effect on firm value through weakening the camouflage effect. When is exogenous, investment is concave in liquidity: Too high liquidity camouflages B s trade, which is capped at. When is endogenous, the camouflage effect is attenuated because rises with liquidity higher ν allows B to trade more, and thus she chooses a higher block. Hence, investment is now monotonically increasing in liquidity. This result echoes Maug (1998), who also shows that liquidity is always desirable when is endogenous. However, our results for exogenous are different. In Maug (1998), if is exogenously high 20, augmenting liquidity reduces firm value (even if liquidity is currently low), and so the benefits of liquidity operate entirely through its effect on initial 19 If q is the proportion of high-quality firms in the economy and r is the probability that a high-quality firm emits s b, then Pr(s b ) = 1 q + qr. Because Pr(X s b ) = qr = 1, r = 1 q and so 1 q+qr 2 q Pr(s b ) = 2 2q. 20 In Maug (1998), liquidity is undesirable if is exogenous and exceeds c M /(H L), where c M is the cost of intervention, and H L is the gain in firm value from intervening. This condition is independent of liquidity, and so if is sufficiently high, raising liquidity can be harmful even if it is initially low. In this paper, raising liquidity from low levels is always beneficial.

19 Blockholder Trading, Market Efficiency, and Managerial Myopia ) block formation. In this paper, increasing liquidity from low levels (if ν< unambiguously increases firm value, even if is exogenous. In sum, there are two effects of greater liquidity. First, as Maug (1998) shows, it leads to larger blocks. Second, it induces increased liquid trading by these blocks. In Maug s model, when is exogenously high, so liquidity only leads to liquid trading, increasing ν is undesirable because it deters blockholder intervention. By contrast, as discussed in Section II.A.1, in this paper liquid trading alone can be beneficial. Hence, liquidity has benefits other than its positive effect on initial block size first shown by Maug. The key to this result is that liquid trading is the very mechanism through which the blockholder adds value, so it is not achieved in intervention models. The result that liquidity can be desirable even when is exogenous is potentially important because legal or institutional factors may deter B from endogenously changing in response to greater liquidity, and thus the first benefit from being obtained (see, for example, Roe (1994)). For example, certain shareholders choose to hold fewer than 5% of a firm s shares to avoid triggering a Section 13(d) filing, or hold fewer than 10% to avoid being classified as an insider. In the United States, ownership is fragmented despite high liquidity, suggesting that these forces may be important. This paper shows that liquidity can be desirable even if it does not lead to Maug s advantage of more concentrated ownership. B. Further Applications In the general model, θ is any action that boosts fundamental value but risks emitting s b. Thus far, θ has been interpreted as intangible investment and s b as short-term losses, but there are many additional applications. Signal s b is any observable characteristic that reduces outsiders assessment of firm value because it is also consistent with a low-quality firm. Therefore, θ can represent fully observable investment for which the motive or quality is unknown. The fundamental problem with investment is that the associated expenditures are difficult to interpret, even if they are fully visible. While R&D can be reported separately on the income statement and atomistic shareholders can costlessly observe it, they do not know whether a rise in R&D results from managerial excess (bad news about agency costs), the need to compensate for failed past R&D efforts (bad news about operating costs), or efficient exploitation of new growth opportunities (good news). Upon observing significant investment for which the motive is unclear, B will gather information and trade accordingly. Low θ can also represent the pursuit of myopic actions that temporarily boost outsiders perceptions, such as accounting manipulation, fraud, or milking customer reputation through lowering product or service quality. Johnson, Ryan, and Tian (2009) document a significant correlation between corporate fraud and unrestricted stock compensation, and Peng and Röell (2008) find that vested options encourage executives to manipulate earnings. Because the manager can sell unrestricted stock and exercise vested options immediately, such compensation increases ω and thus reduces θ. Allowing the manager of a low-quality firm to undertake a value-destructive action that gives a probability

20 2500 The Journal of Finance R of yielding s g (so that s g is also imperfectly informative) would reinforce the results of the core model. The presence of a blockholder reduces the manager s ability to deceive the market about his firm s quality, even in the short run. III. Empirical Implications This paper s broad objective is to demonstrate that blockholders can add value even if they are unable to intervene, which potentially explains the prevalence of small blockholders in the United States. While the model also generates a number of more specific empirical implications, it must be stressed that there are significant challenges in testing them. First, the key variable (block size) is endogenous, as shown in Section II.A.2. Therefore, it is insufficient simply to document significant correlations between block size and an outcome variable. To show that blockholders have the effects predicted in the model, it is necessary to identify sources of exogenous variation in block size. Second, empirical tests will need to take into account the model s specific setting, for example, exclude inside blockholders and blockholders who rarely trade on information (such as families or index funds), and focus on situations where short-sales costs are nontrivial. The implications are divided into three broad themes: the effect of blockholders on financial markets, on firm behavior, and on firm value. I commence with the first category. In this model, blockholders exert governance by being informed traders, which in turn leads to four predictions they have superior information; their trades impound information into market prices; they earn trading profits; and their presence increases market efficiency. Moreover, the magnitude of these effects should be concave in block size. These predictions are unique to a framework where blockholders add value through trading, rather than intervention, and where information acquisition depends endogenously on block size. In standard models of informed trading (e.g., Kyle (1985)), block size has no effect on information acquisition incentives. Blockholders thus generate no excess returns, and their sales are no more informative than any other investor s trades. Owing to the short-sales constraint introduced by this paper, private information is increasing in block size (up to a point). Bushee and Goodman (2007) find that the private information content of an institutional investor s trade is indeed increasing in her stake, and Heflin and Shaw (2000) and Rubin (2007) document that ownership concentration depresses liquidity as other market participants fear informed trading losses. More generally, the paper suggests that ownership concentration may be a more accurate measure of investor informedness than total institutional ownership. The latter plausibly measures the potential to obtain information (because institutions have the expertise to conduct fundamental analysis), but concentrated stakes are necessary to incentivize them to undertake such analysis. If institutional ownership is high but dispersed, shareholders may not bear the costs of monitoring. The second prediction is that, owing to blockholders superior information, their sales should convey negative news and depress the stock price (unless these sales are motivated by noninformational reasons, considered later).